Loss Analysis And Optimization Design Of GaN HEMT Synchronous Rectification Buck Converter

Under the influence of cloud computing and big data,the scale of data centers continues to grow.However,the consequent high energy consumption problem has become one of the key elements that limit the development of data centers.Improving the efficiency of the data center power system,especially the efficiency of the large number of Voltage Regulator Modules(VRMs),is essential to the construction of green data centers.The synchronous rectification Buck converter is widely used in VRMs due to its simple structure and easy-to-control characteristics.As a soft switching solution without auxiliary circuits,the near-Critical Conduction Mode(near-CRM)can improve the response speed while achieving Zero Voltage Switching(ZVS).Because enhancement-Mode Gallium Nitride High Electron Mobility Transistors(e-GaN HEMTs)have the advantages of wide forbidden bandwidth,high switching speed,low on-resistance,the tolerance of high temperature and high pressure,etc.,this paper takes the single-phase near-CRM synchronous rectification Buck converter based on e-GaN HEMTs as the research object,focuses on improving efficiency,and carries out the following researches:Firstly,establish an e-GaN HEMT switching analytical model.On the one hand,consider the loop parasitic inductance,nonlinear junction capacitance and transconductance while establishing the equivalent circuit.On the other hand,combining e-GaN HEMT’s switching characteristics and near-CRM synchronous Buck converter’s large inductor current ripple,the switching transient process is divided into submodes,and the accurate switching analytical model is obtained by numerically solving submodes’ state space equations.Waveform comparison results show that the waveforms obtained from the switching analytical model roughly consistent with the simulation waveforms as well as the experimental waveforms,and the accuracy of the model is relatively high.An accurate switching analytical model can be used to estimate switching losses,which lays the foundation for the following loss analysis.Secondly,the loss distribution of the converter is studied,and key elements related to circuit losses are explored.Based on the switching analytical model as well as the loss calculation methods of inductance and capacitance,the switching loss,inductance loss,capacitance loss and parasitic loss are calculated.Loss analysis results show that the inductance and dead time values have a great influence on the efficiency,and a reasonable selection of these parameters will help improve the efficiency of the converter.Therefore,these parameters need to be well designed in the following parameter optimization design process.Thirdly,taking efficiency as the optimization goal,the inductance and dead time values are designed based on the MATLAB GUI.On the one hand,combining the current ripple requirements and the ZVS boundary conditions,the traversal method is used to determine the range of inductance and dead time.On the other hand,the area product method is used to design the inductor and obtain the parameters required for loss calculations.Based on this,the system loss under each group of parameters is calculated,and the parameters at the highest circuit efficiency are output as the optimization results.Simulation results show that the current ripple requirements and ZVS boundary conditions are met when the optimization results are selected as circuit parameters.Finally,a 12V/3.3V 10 A synchronous Buck experiment platform is built to test the circuit performance under each group of parameters.The experimental results show that the optimized parameters can not only meet the current ripple requirements and ZVS boundary conditions,but also reach the highest system efficiency;after the optimized design process,the system efficiency increases from 73.2% to 79.0%.